Contribution
A five-page review in a themed issue on intergroup relations. It sets out what the minimal group paradigm (MGP) is, what Tajfel’s first studies found, what the critics said, and how the paradigm has been used since about 2011. It has no data of its own and almost no numbers. For the record its use is as a second-hand account of Tajfel et al. 1971, and as a list of the explanations of minimal-group bias that compete with social identity theory.
Key insight
The paradigm was built to remove self-interest, and what is left still favours the in-group. Allocations are made to anonymous others, never to oneself, so “no utilitarian self-interests can be directly served” (p. 85). Social identity theory explained the bias by a motive to “derive meaning and a positive self-concept” from membership (p. 85). The later self-anchoring work gives a different route to the same bias: the new in-group is favoured because the person’s positive view of themselves is projected onto it, “rather than a comparison with the respective outgroup” (p. 86).
Assumptions
- Allocation matrices measure motives. The original design reads choices on Tajfel’s matrices as parity, maximum joint profit, maximum in-group profit or maximum difference (p. 85). Critics said the matrices did not measure these independently (p. 86).
- “Minimal” is a matter of degree. A plausible rationale (overestimator, a painter preferred) differs from an explicitly random one (a coin toss). The first allows stereotypes; the second “undermines meaning making” (p. 87).
- A review’s selection. Otten is a principal of the self-anchoring and positive–negative asymmetry work she reviews (refs. 18, 19, 23–25, 34, 35).
Key results
-
The design (p. 85). Tajfel and collaborators meant to start with a minimal setup and add elements until discrimination appeared. Groups were said to rest on artistic preference (Klee versus Kandinsky). Three features made it “mere categorization”:
- the categorization is novel and arbitrary, with no history;
- it is anonymous, with no face-to-face interaction;
- no utilitarian self-interest is directly served by the allocations.
-
The matrices (p. 85). Participants distributed resources between an anonymous in-group member and an anonymous out-group member. The matrices separated parity, maximum joint profit and maximum in-group profit, and asked whether participants would give up in-group profit for more differentiation. Otten’s illustration: choosing 11 for the in-group and 9 for the out-group over 12 and 11.
-
The first findings (p. 85). “Fairness concerns strongly guided intergroup allocations”, but in-group favouritism emerged reliably. “In some instances” the pull to differentiate was stronger than the pull to maximize in-group profit. This contradicted realistic conflict theory (Sherif 1967). It inspired social identity theory (Tajfel & Turner 1979) and self-categorization theory (Turner et al. 1987).
-
Criticism (p. 86).
- The matrices were said to confound strategies (Bornstein et al. 1983). The findings replicated with new matrices.
- The artificial setting may give category information “unrealistic attention” (Diehl 1989).
- Reciprocity expectations may make the bias utilitarian after all (Gaertner & Insko 2000).
- Positive–negative asymmetry: bias is mostly confined to allocating positive resources and positive traits. With punishment or negative traits, “‘mere categorization’ has no longer a reliable impact”.
- Self-anchoring: favouritism may come from projection of the positive self onto the minimal in-group (Cadinu & Rothbart 1996; Gramzow and colleagues; Otten 2002).
Otten’s verdict: the MGP “has remained a reliable and useful instrument”.
-
Automatic responses (pp. 86–87). Right after assignment, people have more positive implicit associations with the new in-group. In-group faces are encoded faster (Ratner & Amodio 2013), but the face-memory bias held for European Canadians and not East Asians (Ng et al. 2016). Americans favoured a minimal in-group more than Japanese did, “partly, but not fully explained” by self-esteem (Falk et al. 2014). Out-group faces were more readily encoded as angry, for minimal as for racial groups (Dunham 2011). Products were better built from instructions said to come from an in-group member, and the gap shrank when a superordinate category was introduced (Greenaway et al. 2015). Preschoolers did not trust a reliable out-group informant over an unreliable in-group one (MacDonald et al. 2013). Physiological threat arose when a minimal group’s status was threatened (Scheepers et al. 2009).
-
Two directions of self–group alignment (p. 87). With an anonymous, undefined group, identification followed how far people projected their own traits onto it (self-anchoring). Once stereotypes about the group were given, projection from group to self (self-stereotyping) predicted identification best (van Veelen et al. 2013, 2016).
-
Induction procedures (p. 87). Pinter and Greenwald (2011) found similar explicit favouritism across four procedures, and the largest implicit effects when participants only memorized one group’s names. Otten reports the authors’ reading that this links the self-concept to the studied group.
-
Self-interest included (pp. 87–88). Harvey and Bourhis (2012, 2013) added a Monopoly game, group wealth and wealth stratification.
- Poor groups favoured the in-group whether wealth came by chance or merit. The rich favoured it strongly only if their wealth was earned.
- Dividing money between self, in-group and out-group, “participants gave most money to themselves, but kept favoring their ingroup over the outgroup in their distributions of the remaining resources” (p. 87).
-
Conclusion (p. 88). The paradigm is now used less to understand discrimination than to study “a whole range of other interpersonal, intra-group and inter-group phenomena”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Novel, arbitrary, anonymous categorization, with no personal stake in the allocations, reliably produces in-group favouritism, alongside a strong pull towards fairness | moderate as a review: asserted from Tajfel et al. 1971, Brewer 1979 and Hewstone 2002, with no effect sizes | pp. 85–86 |
| C2 | Seeking positive differentiation sometimes outweighs maximizing the in-group’s absolute gain | weak to moderate: “in some instances”, no figures | p. 85 |
| C3 | Minimal-group bias is mostly confined to allocating positive resources and attributing positive traits | moderate: the author’s own programme | p. 86 |
| C4 | A link between the individual self and the novel in-group, by projection, is a crucial factor in minimal-group bias | moderate: several cited studies, most from the author’s circle | pp. 86–87 |
| C5 | When people can allocate to themselves, they take the most for themselves and still favour the in-group with the remainder | weak to moderate: two cited studies from one lab, no figures | pp. 87–88 |
Concepts
- minimal group paradigm: categorization that is novel, arbitrary and anonymous, with no instrumental link between self-interest and the allocations or evaluations (p. 85).
- maximum differentiation: giving up in-group gain for a larger in-group advantage over the out-group.
- positive–negative asymmetry: minimal-group bias appears with positive outcomes and largely not with negative ones.
- self-anchoring: projecting one’s own traits onto a group, so that the group shares the self’s (usually positive) evaluation.
- self-stereotyping: projecting the group’s traits onto oneself.
Connections
- Tajfel et al. 1971 (LIT-1697), Unavailable. This review gives its design and the gist of its result, not its experiments or numbers.
- Robertson et al. (NOTE-1245). Their preregistered online study found minimal-group favouritism only as it rose with identification. That fits Otten’s emphasis on the self–group link as the driver (C4). It is not the same claim: Robertson measured identification, not projection.
- Hornsey 2008 (NOTE-1253) names the same critiques at the same level of detail: demand, similarity and Rabbie’s interdependence. Otten adds positive–negative asymmetry and self-anchoring, which Hornsey does not.
- Self-categorization theory (THEORY-1114). Van Veelen’s self-stereotyping is the theory’s depersonalization seen from the individual. Self-anchoring runs the other way. Otten treats the two as complementary pathways chosen by how much is known about the group.
- Self-esteem (LIT-1780, unread). Falk et al.’s cultural result, partly explained by self-esteem, is one more piece of the contested self-esteem strand.
Bearing on the record
-
The owner’s thesis (CLAIM-491). The paper bears on it at three points. These are inferences; Otten does not discuss selflessness.
- The minimal group is the textbook case of acting for a group with no personal gain. Its standard explanation, a positive self-concept from membership, already has the form of the thesis: a self-interest served through the group.
- Self-anchoring and self-stereotyping are two routes to the same identification. Self-anchoring extends the personal self over the group: the group is valued because it carries the self. That fits a “larger self”. Self-stereotyping fits SCT’s “changed self”, which the claim’s “What it does not say” flags as the poorer fit. On Otten’s account, which route operates depends on how much the person knows about the group.
- When the personal self can be paid directly, it is paid first. The group is favoured with what is left. The extended self then has an order inside it, with the personal self at the front. The thesis says nothing yet about that order.
-
The triviality objection (CLAIM-492). Projection of one’s own traits is a criterion of the self’s extent that is not a report of caring. Otten reports it predicts identification and bias. It is measured by trait overlap between self and group, the same kind of measure as Aron’s circles.
-
Batson’s objection (CLAIM-494). Not addressed.
-
A dated note should be added to CLAIM-491, perhaps together with Pinter and Greenwald’s (see the second draft). Proposed text:
Note of 2026-10-29: the minimal group, read at second hand. Otten (2016) reports two routes from a new self-category to in-group favouritism: projecting the positive personal self onto the group (self-anchoring), and taking on the group’s traits (self-stereotyping). The first is the thesis’s larger self and the second its changed self, and which operates depends on how much is known about the group. When participants could also pay themselves, they took the most for themselves and favoured the in-group with the rest, so the extended self, if that is what this is, puts the personal self first.
-
No THEORY filed. A review with no data, of a paradigm the record holds at second hand.
-
No ML instruction.
Limitations
- Not the original. No sample, no matrix, no pull score, no experiment from Tajfel et al. 1971 is given. The reader cannot tell from this whether the 1971 paper reported the “maximum differentiation” result itself or whether it came later.
- A short review with few numbers. No effect size is given for any study.
- The author’s own work carries the positive–negative asymmetry and self-anchoring sections.
- An internal inconsistency. The text says Falk et al.’s cultural difference was “partly, but not fully explained” by self-esteem (p. 86). The annotated reference says it “could mostly be traced back to” self-esteem (p. 88).
- Typos.
- “MPG-experiments” for “MGP-experiments” (p. 86).
- “This effects of variations in how minimal social categories are was examined” (p. 87).
- “preferring one painter of the other” for “or the other” (p. 87).
- “an adopted version” for “adapted” (annotation to ref. 15, p. 88).
- Ref. 37, Pinter & Greenwald, Group Process Interg, is dated 2001; its DOI and volume (7) are 2004’s (p. 89).
Open questions
- What did Tajfel et al. 1971 actually report: how large was favouritism, how strong was parity, and was maximum differentiation shown there or later? Only the original, or a reanalysis of it, can answer.
- Does self-anchoring predict sacrifice for the group when caring about the group is held fixed? That would test it as the criterion CLAIM-492 asks for.